Online reporting: made readable, usable and citable for LLMs

What happens when your key stakeholders no longer read your annual or sustainability report but ask ChatGPT to explain it instead? Then what matters is whether AI can reliably find your content, understand it correctly and cite it accurately. This is why online reporting will become indispensable for the future of financial reporting – with HTML as the central format.

When people search for information online today, the trend is to bypass traditional search engines in favour of assistants like ChatGPT that are powered by large language models (LLMs). This has direct implications for the design of publishing strategies in corporate communications.

The reason lies in the fundamental difference in the way they operate. Search engines work with keywords, links and rankings. LLMs, by contrast, interpret content semantically and contextually, generating answers from structured, machine-readable sources. Companies will therefore need to go beyond classic SEO and focus on Generative Engine Optimization (GEO), as LLMs need content that is not just well written, but unambiguously structured.

This development is particularly relevant for financial reporting. Financial reports are not only labour- and cost-intensive but also enjoy high credibility thanks to internal and external auditing. For both reasons, it makes sense to prepare them in a way that allows them to be used optimally in LLM-based information retrieval.

HTML does more than display content: it structures it so that ChatGPT and similar systems can read it reliably and also interpret and reproduce it correctly. Texts, tables, KPIs, charts and relationships are semantically tagged, making them machine-interpretable. This enables connections – including hierarchies, KPI linkages, time periods and context – to become instantly clear.

Studies show that LLM-based systems cite structured sources more than 2.3 times as often, whereas with PDFs, the systems are more likely to draw on references from external sources such as media reports – with corresponding risks of error and reputational damage. HTML content is therefore not only found more quickly, but also reproduced more frequently and more consistently. This means that reports are transformed into digital knowledge bases – with verified content that’s machine-readable, interactive, shareable and measurable.

On top of this, online reports add clear communicative value. Interactive dashboards, pop-up glossaries, animations or videos from senior management make reports more accessible, increase time spent on the page and create tangible value for readers.

As the primary reporting instrument, the PDF is reaching its limits in terms of discoverability, as language models cannot reliably interpret its structure and semantic relationships. As a result, content is often only partially understood and utilised. That said, the PDF remains relevant as a legally binding reference and audit-proof archive document.

What’s needed is reporting that delivers content in a consistent and channel-appropriate way: while the PDF remains the audit-proof reference, HTML maximises communicative impact and supports machine use. Companies that intelligently combine both formats and address a range of relevant touchpoints (such as websites, investor relations, social media and newsletters) not only meet regulatory requirements but also strengthen their impact – even as information is increasingly conveyed via ChatGPT and other AI assistants.

The most efficient approach is a platform-based model with a central source of data and content. Content is created, reviewed and approved once, then automatically published across all the required formats: the online report (HTML), the PDF and, thanks to integrated tagging, XBRL or iXBRL. This eliminates time-consuming manual input across multiple systems, making processes faster, more consistent and significantly less error-prone.

The direction of travel is clear: towards integrated platforms and full HTML. Companies that manage content centrally and distribute it automatically gain efficiency while retaining full control over verified statements and key metrics. We would be happy to help you redesign your reporting to make it fit for the AI transformation – with strategies that combine compliance, processes and credibility.